[SmartCat]

Originally Posted by balenoed_

Why is it taking so much time to find the CVR? Considering it is not one of the remotest ocean on the earth like where MH370 was doomed. Also there is a location transmitter beacon which is active for 30 days which might help to narrow down the search area.

Because of the speed with which the aircraft hit the water, cockpit voice recorder is likely to be a couple of feet inside the seabed. The signals from voice recorder is very faint.

https://www.thehindubusinessline.com...le25418422.ece

[Jeroen]

Originally Posted by Hayek

. A defective product that kills even 300 people and injures 20000 a year would drive its makers bankrupt. It’s only if self driving cars are 10000x better (3 deaths and 200 injuries a year) that they can be introduced without new laws that protect manufacturers from tort claims. The same applies to pilotless planes.

True. For some reason societies the world are very sceptical about self driving cars and expect such a car to be much safer than a human driven car.

We see a lot of very theoretical debate these days on self driving cars and how they would decided in certain scenario’s. Scenario’s are likely to be along the following lines:

The car is driving along a road, normal appropiate speed, all of a sudden a little old lady crossed the road from the left and a young girl from the right. There is no way the car can come to a stop, it needs to divert to either right or left to hit only one person, staying in the middle will kill them both. How should the computer respond.

Interestingly enough humans driver are not required to think through such a scenario. They are not taught anything remotely as such, neither is it tested during the driving test.

Also, we know that most driver absolutely such at making any such real emergency decision. But we still believe a self driven car needs to outperform by a very large margin anything we expect from a human driver.

Beats me.

Originally Posted by balenoed_

Why is it taking so much time to find the CVR? Considering it is not one of the remotest ocean on the earth like where MH370 was doomed. Also there is a location transmitter beacon which is active for 30 days which might help to narrow down the search area.
.

Apparently, they lost the CVR location signal a few days ago. Which is going to make it more difficult to find the CVR that is likely to be stuck under a thick layer of mud.

Jeroen

[SmartCat]

Originally Posted by Jeroen

The car is driving along a road, normal appropiate speed, all of a sudden a little old lady crossed the road from the left and a young girl from the right. There is no way the car can come to a stop, it needs to divert to either right or left to hit only one person, staying in the middle will kill them both. How should the computer respond.

If sensors detect people on either side of the road, then it should slow down further - and not continue along at "normal appropriate speed". That is, if the computer cannot bring the car to a halt (or slow down enough) from 60 kmph, then it should reduce speed to 30 kmph. So that impact speed is 10 kmph, assuming the 2 people continue along their path and the computer is forced to slam the brakes. The system should incorporate pretty much all possibilities.

After all, you cannot assume that living things behave rationally. What if there are cows on either side of the road?

[Jeroen]

Originally Posted by smartcat

If the sensors detect people on both the sides of the road, then it should slow down further - and not continue along at "normal appropriate speed". That is, if the computer cannot slow down enough from 60 kmph, then it should reduce speed to 30 kmph. The system should incorporate pretty much all possibilities.

Maybe I did not explain the scenario in sufficient detail. The debate is about how self driven cars, i.e. computers take life/death decisions. In this scenario, whether possible or not, the old lady and girl just materialised on the road and even by applying maximum brake the car would not be able to slow down sufficiently. So it needs to choose between hitting both the old lady and the girl, or hitting only the old lady or the girl.

Their are endless varieties to the theme. A group of children versus a a group of pensioners.

The idea is that self driving car will face situations with a moral dilemma. Do I run over the little old lady who is at the end of her life, or the young girl that still has a whole life in front of her.

Apart from the technological challenges of a computer determining a young girl from an old woman, it is of course possible to program a certain “moral compass” behaviour into a computer. But how do you determine which is the best choice?

Let alone, to my earlier points, human drivers are pretty awful to start with at making these split second decisions in the first place.

Jeroen

[SmartCat]

Originally Posted by Jeroen

So it needs to choose between hitting both the old lady and the girl, or hitting only the old lady or the girl. A group of children versus a a group of pensioners. The idea is that self driving car will face situations with a moral dilemma. Do I run over the little old lady who is at the end of her life, or the young girl that still has a whole life in front of her.

Apart from the technological challenges of a computer determining a young girl from an old woman, it is of course possible to program a certain “moral compass” behaviour into a computer. But how do you determine which is the best choice?

Oh man, this is an unnecessary complication. The best choice is to slam the brakes and take the hit. That's what rational humans drivers do in such a situation anyway. Put yourself in the shoes of such a driver -> can you choose between an old lady and a child? You won't have time to choose between your "targets" anyway.

[Jeroen]

Originally Posted by smartcat

Oh man, this is an unnecessary complication. The best choice is to slam the brakes and take the hit. That's what rational humans drivers do in such a situation anyway. Put yourself in the shoes of such a driver -> can you choose between an old lady and a child? You won't have time to choose between your "targets" anyway.

No, your choice is your choice and not necessarily the best. You seem to opt for hitting both. Of course, in these scenario’s there simply is no good or bad choice. it depends on one’s own moral compass.

Also, drivers do not make rational decision. Plenty of accidents where the drivers freezes up, don’t do anything or do something completely stupid.

The point being, for some reason society likes to see a self driving car outperform a human driver by a very large margin. We do not require human drivers to think this through and act accordingly.

Jeroen

[Jeroen]

For those who would like to experience first hand what I was trying to explain, have a go at this:

http://moralmachine.mit.edu

Make your own judgements and compare your results to others.

Than try and figure out how we are going to program our self driving cars.

Jeroen

[balenoed_]

Picture and article courtesy:
https://www.seattletimes.com/busines...ion-air-crash/

Three former Boeing flight control experts were startled by the FAA’s description last week of the new MAX system. In an airworthiness directive, the FAA cited an analysis by Boeing that “if an erroneously high single angle of attack (AOA) sensor input is received by the flight control system, there is a potential for repeated nose-down trim commands” that will swivel the plane’s horizontal tail to pitch the nose downward.

The fact that the plane’s nose could be automatically and repeatedly pushed down due to one false signal shocked Peter Lemme, a former Boeing flight controls engineer, who said it looks like a design flaw.

“To contemplate commanding the (horizontal tail to pitch the jet) nose down clearly is a major concern. For it to have been triggered by something as small as a sensor error is staggering,” Lemme said. “It means somebody didn’t do their job. There’s going to be hell to pay for that.”

[balenoed_]

The same explained by a captain in a video.

http://www.youtube.com/watch?v=zfQW0upkVus

[balenoed_]

I have a question here:

Even though the AoA sensor was providing erroneous data, why would the MCAS interpret it as a stall situation when the aircraft had some good enough speed (200-300 knots ?) provided the speed sensors are giving reliable speeds.

I may be totally wrong in understanding the facts, please excuse if this does not make sense.

[Jeroen]

Originally Posted by balenoed_

Even though the AoA sensor was providing erroneous data, why would the MCAS interpret it as a stall situation when the aircraft had some good enough speed (200-300 knots ?) provided the speed sensors are giving reliable speeds.
.

I really don’t know. We need to wait on the accident investigation report. Although I imagine there will be a few more updates from Boeing and or the investigation body.

But a wonky AoA sensor gives false inputs to the flight management and control system. We don’t know if the pilots had reliable air speed when this happened. Airspeed alone doesn’t mean much either. You can stall any plane any time at any air speed. You might be perfectly safe flying straight and level at 200 knots indicated airspeed. But when you pull a sharp turn left or right, or pull the nose up you might find yourself in a stall!

But even if they did, if the AoA input suggests that you are in stall, the MCAS will push the nose down. That’s how I understand it. Next, it wasn’t quite clear how to override it and get it secured manually. To add to the confusion, it appeared as an intermittent problem and the pilots were probably faced with a multitude of different alarms and caution warnings.

Jeroen

[balenoed_]

Originally Posted by Jeroen

But even if they did, if the AoA input suggests that you are in stall, the MCAS will push the nose down. That’s how I understand it. Next, it wasn’t quite clear how to override it and get it secured manually.

The discussions in the aviation forums and what Boeing says is that the pilots should control the trim - the angle of the stabilizers to counteract to what the MCAS was doing to the stabilizers. That is only a temporary solution though, until the MCAS will kick in and push the nose down a few seconds later once they release it.

The crucial step, according to the Boeing bulletin, would be to flip a pair of switches off. The switches which disables the electric control of the motor that moves the stabilizers up and down, preventing the anti-stall system from exerting control over their position.

[Jeroen]

Originally Posted by balenoed_

The discussions in the aviation forums and what Boeing says is that the pilots should control the trim - the angle of the stabilizers to counteract to what the MCAS was doing to the stabilizers. That is only a temporary solution though, until the MCAS will kick in and push the nose down a few seconds later once they release it.

The crucial step, according to the Boeing bulletin, would be to flip a pair of switches off. The switches which disables the electric control of the motor that moves the stabilizers up and down, preventing the anti-stall system from exerting control over their position.

The problem was as I mentioned intermittent. A natural correct response of the pilot would be to push the appropiate trim button. That would actually give the desired response. But when he released the trim a few seconds later the MCAS would push the nose down again. Also, the way I read the various discussion is that you could only disable the MACS after you had re-trimmed, before it kicked back in.

Jeroen

[Rahul Bhalgat]

One general question, not specifically for this accident only:

It sometimes becomes costly and impossible to retrieve the CVR and FDR from the ocean floor. Why not design them such that they will detach and float upon impact / disintegration?

[itwasntme]

Just spotted SpiceJet's brand new VT-MXE (christened 'Tarragon') at Mumbai (CSIA) pushing back for Cochin.

Looks like SG pressed it into service within a couple of days of it arriving from Seattle via Manchester. Hoping MCAS has been added to flight manuals and also practiced by the pilots.

FR24:
https://www.flightradar24.com/data/aircraft/vt-mxe

Planespotters:
https://www.planespotters.net/airfra...ceJet/a19OugrJ